Cycle length variability (CLV), defined as the standard deviation of normal cycle length intervals, has been found to be a powerful predictor of subsequent mortality in a population of 808 survivors of acute myocardial infarction. Decreased CLV is associated with a significant increase in mortality. CLV remained an independent predictor of outcome even after adjusting for left ventricular ejection fraction, clinical risk factors, heart rate and ventricular arrhythmias. In the same population of survivors of acute myocardial infarction, the results of exercise testing also strongly predicted outcome, with those failing to take the test having the worst survival, and those completing the low-level stress test taken before discharge having the best prognosis. The hypothesis that the status of stress test (completed; did not complete; failed to take) and CLV were measuring the same factor related to mortality was tested. Although the distribution of CLV was shifted to higher CLV in patients who completed the test and to lower CLV in those who failed to take the test, both predictors of mortality remained independent predictors of long-term mortality (average of 31 months of follow-up) after controlling for each other. Moreover, subgroups with an approximate 15-fold difference in mortality were defined using both variables (CLV < 50 ms, did not take test had a 54% mortality; CLV > 100 ms, completed the test had a mortality of 3.5%). CLV is a measure of autonomic tone; it is not strongly related to exercise ability and using the results of both stress testing and CLV results in the identification of subgroups of postinfarction patients with markedly disparate risks of mortality.